Obscene image detection refers to the process of identifying obscene and pornographic portions in images previously extracted from a given video file; a process that constitutes the core of the broader obscene-video filtering system. Existing obscene-image detection methods rely on information about image texture such as RGB proportions, color-distribution histograms, and YIG to track the skin-color and edge information of the image concerned. Existing methods, however, are not very accurate when it comes to determining the obscenity level in low-quality UCC videos. This paper proposes an improved method that first utilizes Canny Edge to analyze the fine grains of the image to determine whether the image is of high or low quality, and then employs to determine whether the image passes the final obscenity test. In order to check for the efficacy of this method, an arbitrarily selected batch of images was first put through the Canny Edge test to separate the batch into two groups based on the image-quality level. The images were then tested for their obscenity levels twice, first with an existing method and then with the method proposed in this paper. Results were then analyzed, which showed that the new method yielded results that were about 10% more accurate.
[1]
David A. Forsyth,et al.
Finding Naked People
,
1996,
ECCV.
[2]
Chang-Hsing Lee,et al.
An adult image identification system employing image retrieval technique
,
2007,
Pattern Recognit. Lett..
[3]
Chung-Lin Huang,et al.
Hand gesture recognition using a real-time tracking method and hidden Markov models
,
2003,
Image Vis. Comput..
[4]
Milind R. Naphade,et al.
A probabilistic framework for semantic video indexing, filtering, and retrieval
,
2001,
IEEE Trans. Multim..
[5]
Paul A. Watters,et al.
Identifying and Blocking Pornographic Content
,
2005,
21st International Conference on Data Engineering Workshops (ICDEW'05).
[6]
Kyoungro Yoon,et al.
Multi Class Adult Image Classification Using Neural Networks
,
2005,
Canadian Conference on AI.
[7]
Shohreh Kasaei,et al.
Comprehensive Evaluation of the Pixel – Based Skin Detection Approach for Pornography Filtering in the Internet Resources
,
2005
.